14–15 Nov 2026
Bengaluru
Asia/Kolkata timezone

Agentic Edge AI on Ubuntu Core: Building Autonomous Industrial IoT Systems with Raspberry Pi 5

Not scheduled
1h
Workshop Room

Workshop Room

Workshop (60 mins) Embedded Systems & IoT

Speaker

Mr Saravanan Ganesan

Description

In this workshop, we will explore how Ubuntu Core and Raspberry Pi 5 can be leveraged to build secure, resilient, and production-oriented Agentic AI systems for Industrial IoT. The session focuses on moving beyond traditional edge monitoring toward autonomous systems that can monitor industrial assets, detect anomalies, understand operational context, reason over the available data, and initiate controlled corrective actions locally.

Attendees will learn how open-source technologies can be combined to build an intelligent edge architecture using Ubuntu Core, MQTT, Edge AI models, Ollama, Microsoft Phi-3 Small, AI Agents, OPC-UA, and Modbus. The architecture is designed with local processing in mind, allowing the core monitoring and AI decision-making workflow to continue even when cloud connectivity is unavailable.

The workshop will include a live Raspberry Pi 5 smart-factory demonstration that brings these components together. Sensor and equipment telemetry will be collected through MQTT and representative industrial protocols such as OPC-UA and Modbus. A lightweight Edge AI component will analyse the telemetry and identify abnormal equipment behaviour. The relevant telemetry and operational context will then be passed to a locally running AI agent.

The AI agent, powered by Ollama and Microsoft's Phi-3 Small language model, will interpret the available information, reason about the potential equipment condition, explain its decision, generate recommendations, and use predefined tools to trigger controlled workflows when appropriate. This demonstrates a complete Sense → Detect → Understand → Reason → Decide → Act loop running directly on the edge device.

Demo Technology Stack

The live demonstration will use:

Raspberry Pi 5 – The primary edge device running the AI and Industrial IoT workloads locally.
Ubuntu Core – Provides a secure, lightweight, and production-oriented foundation for the edge device.
MQTT – Used for real-time sensor and equipment telemetry exchange.
OPC-UA & Modbus – Used to represent connectivity with industrial equipment, PLCs, and factory-floor systems.
Edge AI / Anomaly Detection – Identifies abnormal patterns in equipment and sensor telemetry.
Ollama – Provides the local LLM runtime on the Raspberry Pi.
Microsoft Phi-3 Small – Provides local language understanding and reasoning capabilities.
AI Agent – A Python-based agent that coordinates telemetry, anomaly information, operational context, reasoning, and tool execution.
Agent Tools & Workflows – Predefined tools allow the agent to perform controlled actions such as generating alerts, updating equipment status, or triggering corrective workflows.
Local Data & Context – Equipment information, sensor readings, and operational conditions are supplied to the agent for decision-making.
Disconnected Operation – The core monitoring and reasoning pipeline operates locally without requiring continuous cloud connectivity.

The session will also cover the practical challenges of bringing Agentic AI to constrained edge devices, including resource limitations, local model inference, latency, resilience, security, controlled tool execution, and cloud dependency. The goal is not simply to demonstrate an LLM running on a Raspberry Pi, but to show how an AI agent can become part of an actual industrial monitoring and automation loop while maintaining appropriate boundaries around autonomous actions.

By the end of the workshop, attendees will have a practical understanding of where Agentic AI fits within an Industrial Edge architecture, how Ubuntu Core can provide a secure and maintainable platform for deploying these workloads, and how open-source technologies can be combined to build autonomous edge solutions for predictive maintenance, equipment monitoring, intelligent automation, and Industry 5.0 use cases.

The workshop is designed for developers, IoT engineers, AI/ML practitioners, and technology enthusiasts interested in understanding how AI, industrial protocols, and edge computing can come together to build the next generation of autonomous industrial systems.

Summary

The demo showcases an autonomous industrial AI agent running entirely on a Raspberry Pi 5 using Ubuntu Core, Ollama, and Microsoft's Phi-3 Small Language Model, enabling real-time reasoning and decision-making at the edge without cloud dependency.

Speaker Bio

Saravanan Ganesan is a Microsoft MVP in Azure AI and Azure IoT, and an Assistant Consultant at Tata Consultancy Services (TCS). With over 12 years of experience building intelligent cloud-to-edge solutions, he specializes in Edge AI, Industrial IoT, Digital Twins, AI-powered automation, and large-scale cloud platforms using Azure and open-source technologies. He is an active community speaker, mentor, and technical organizer who regularly shares knowledge through developer communities, workshops, hackathons, and global technology events.

Prerequisites

No prior experience with Agentic AI is required. A basic knowledge of Ubuntu, Raspberry Pi, or IoT systems will help attendees get the most out of this session.

Difficulty Intermediate
Speaker LinkedIn / Socials https://www.linkedin.com/in/saravanan-ganesan-23705a47

Presentation materials

There are no materials yet.